The AI world is getting ‘loopy’

2026-06-22 · Source: TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Robotics & Autonomous Systems · Depth: Advanced, extended

Summary

Boris Cherny, creator of Claude Code, highlighted the emergence of AI "loops" as a significant advancement in agentic AI at Meta's @Scale conference. These loops involve agents continuously prompting other agents to perform tasks like improving code architecture or unifying duplicated abstractions, submitting pull requests, and running endlessly. Cherny noted that 30% of his code is now written by loops, and Anthropic has seen an 8x increase in code per engineer. While token-intensive and potentially costly, the focus shifts to return on investment (ROI) rather than just expenditure. This approach extends to automated code review, security scanning, and even optimizing CI times by 50% using dynamic workflows and millions of tokens.

Key takeaway

For AI Engineers and MLOps teams evaluating advanced automation, embrace agentic loops and dynamic workflows despite their higher token consumption. Focus on the substantial ROI from continuous self-improvement in areas like code maintenance, security, and CI optimization, rather than solely on immediate costs. Implement "auto mode" for agent permissions and "exploratory" output styles to maintain oversight and foster learning while scaling autonomous operations.

Key insights

AI agents continuously prompt other agents in loops for autonomous, self-improving software development and operational tasks.

Principles

• Prioritize ROI over raw cost in AI adoption.
• Embrace continuous, non-deterministic agentic processes.
• Utilize test-time compute for superior outcomes.

Method

Agentic loops involve one agent continually seeking improvements (e.g., code architecture, duplicate abstractions) while another executes, submitting changes like pull requests. Non-deterministic logic dictates when loops stop. Dynamic workflows orchestrate sub-agents.

In practice

• Deploy agents to continuously improve code architecture.
• Automate project management tasks like status updates.
• Use AI for autonomous travel booking and itinerary management.

Topics

• Agentic AI
• AI Loops
• Claude Code
• Dynamic Workflows
• Code Automation
• MLOps
• Test-Time Compute

Best for: AI Architect, CTO, VP of Engineering/Data, AI Engineer, Machine Learning Engineer, MLOps Engineer

Related on AIssential

• I guess we're writing loops now? — Agents can dynamically orchestrate complex, multi-stage coding tasks through self-prompting loops, automating entire development workflows.
• Claude Code, Codex and Agentic Coding #7: Auto Mode — Auto Mode enhances agentic coding safety by selectively requiring human approval for high-risk actions.
• Only the best are using them... — AI coding loops automate development by defining a verifiable end goal, allowing agents to self-iterate until completion.
• Most Enterprise Agentic Projects Are Doomed, Here's Why — Jess Grogan-Avignon & Jack Wang, Accenture — Enterprise AI success requires adapting organizational speed, finance, delivery, and trust-building to AI's emergent, machine-speed nature.
• This Week in AI: Fable 5, the Clone Wave, and Uber’s AI Reality Check — Efficient AI agent development hinges on leveraging existing "ingredients" and building agent-native, observable tooling to manage costs.
• Loop Engineering — Design autonomous systems ("loops") to prompt coding agents, replacing direct human interaction for iterative task completion.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by TechCrunch.